Step shift: A fast image segmentation algorithm and its hardware implementation for next-generation sequencing fluorescence data

摘要

In this study, an image segmentation algorithm for processing Next-Generation Sequencing (NGS) fluorescence data is proposed. In NGS applications, fluorescent spots are randomly distributed, the number of spots is large, and the sizes of these spots are small. All the characteristics make the conventional segmentation algorithm not as efficient. To speed up the segmentation process, we propose a hardware-compatible algorithm, Step Shift, and implement an ASIC design for hardware acceleration. When compared to the software version of Step Shift method, our hardware accelerator can reach 9.2 times speed-up in the experiments.